The mysterious decline of home advantage

This article shows that home advantage in European club football is massively decreasing since the seventies.
When I started this site, the oldest results that were included in the database were from 1993. I used them mainly for calibration and to have evened out Elo values for today's clubs. I discovered that on average, the home advantage is approximately 90 Elo points (measured on a 4-year period between 2006 and 2010). Have a look at this article for details.
As soon as I included historic results dating back to the fifties and beyond, something strange happened. On average, home teams won signifiacntly more Elo points than expected. After having carefully checked for calculation errors, I decided not to have a look at the average of the whole database but to investigate how this phenomenon developed over time. The following graph shows how many points were exchanged per match for each year. The graph is slightly smoothend, positive values indicate that home teams won more Elo points than away teams.

Average points exchanged+10-1Year196019701980199020002010

This is not what I wanted. For a match of football, both teams should have the same probabilities to win or lose Elo points and this is not the case here as home teams on average take up to 1.1 points from the away teams per game.
On the other hand, you can clearly see how home advantage changed over time. I do not want to overinterprete this chart, even though there is a huge amount of games in the database. However, there seems to be a maximum of bias towards home teams in the seventies. Since then, the home advantage constantly decreases until today. You can see, that for the late 2000s, a home advantage of 90 Elo points was accurate. This chart only shows a trend and it does not yet directly show how massive this change is.
I decided to compensate for this phenomenon in my formulas and not to fix the home advantage at 90 anymore but to make it time-variable. After trying out a few parameters, I set the home advantage to be 150 Elo points in 1985 and before and to 90 from 2005 onwards. In between, the home advantage decreases by 3 points a year. In European cups, experience shows that the home advantage is bigger, it is now set at 240 Elo points before 1985, 120 Elo points since 2005 and gradually decreasing inbetween.
Home advantage in Elo points+2001000Year196019701980199020002010New method (Eurocup)New method (domestic)Old method

With these new adjustments for home advantage, home teams roughly have the same probabilities as away teams to gain Elo points in every era as you can see in the following chart:
Average Elo points exchanged+10-1Year196019701980199020002010New methodOld method

So what does this mean? Elo ratings of clubs represent their estimated strength at a point in time. When 2 teams played each other in the seventies or before, the game was completely open when the away team was 150 Elo points better than the home team. Today, if the away team is just more than 90 Elo points better than the home team, it is considered the favourite.
Since the seventies, the home advantage for a European club football match has almost halved and this trend is still continuing today. To illustrate this: If the current trend goes on, there will not be any home advantage anymore by 2025-2030. I do not expect that to happen, this is just to display the magnitude of this phenomenon.
One question remains: Why? There are many possible reasons for this, better infrastructure and away game preparation, different tactics, Three-points-for-a-win to name a few. This Guardian article is worth a read in this matter.

Created: 23 Mar 2013 - Modified: 14 Jul 2012